Skip to main content
NCBI home page
International Journal of Methods in Psychiatric Research logoLink to International Journal of Methods in Psychiatric Research
. 2014 Feb 5;23(2):245–256. doi: 10.1002/mpr.1424

Profiling the subjective effects of Δ9‐tetrahydrocannabinol using visual analogue scales

Daniël Kleinloog 1,✉, Frits Roozen 1, Willem De Winter 1, Jan Freijer 1, Joop Van Gerven 1
PMCID: PMC6878449  PMID: 24496889

Abstract

The subjective effects of cannabis and its main psychoactive component Δ9‐tetrahydrocannabinol (THC) have played an important part in determining the therapeutic potential of cannabinoid agonists and antagonists. The effects mainly consist of feeling high, changes in perception, feelings of relaxation and occasionally dysphoric reactions. These effects are captured by two of the most frequently used visual analogue scales (VASs) in clinical (pharmacologic) research to measure subjective effects: VAS Bond and Lader (alertness, calmness and mood) and VAS Bowdle (psychedelic effects). In this analysis, the effects of THC on these VASs were compared within a total of 217 subjects who participated in 10 different studies. Not surprisingly, the item feeling high was found to be the best predictor for the effect of THC. Three separate clusters that describe the spectrum of subjective effects of THC were identified using different statistical methods, consisting of VAS “time”, “thoughts” and “high” (“perception”), VAS “drowsy”, “muzzy”, “mentally slow” and “dreamy” (“relaxation”) and VAS “voices”, “meaning” and “suspicious” (“dysphoria”). These results provide experimental evidence that THC can evoke different classes of effects. These distinct subjective clusters could represent effects on various systems in the brain, which can be used to further differentiate the involvement of endocannabinoid systems in health and disease. Copyright © 2014 John Wiley & Sons, Ltd.

Keywords: visual analogue scales, VAS, THC, subjective, profile

Introduction

Cannabis is best known as a recreational drug that is widely used throughout the world, although its use is illegal in most countries. The main psychoactive component of cannabis is Δ9‐tetrahydrocannabinol (THC), which is an agonist at cannabinoid CB1 and CB2 receptors. Both cannabis and THC have been used in preclinical and clinical research to investigate the effects on pain (reviewed by Lynch and Campbell, 2011), food intake (reviewed by Berry and Mechoulam, 2002), pain and spasticity in multiple sclerosis (reviewed by Karst et al., 2010; Zajicek and Apostu, 2011) and as models of psychosis (D'Souza et al., 2004; Kleinloog et al., 2012; Liem‐Moolenaar et al., 2010). Alternatively, CB1‐antagonists (e.g. rimonabant) have been shown to block the anti‐nociceptive effects of THC (Compton et al., 1996), reduce food intake and facilitate weight loss (van Gaal et al., 2005), and there is literature that suggests a relation between CB1‐antagonism and the development of multiple sclerosis (van Oosten et al., 2004) and improvement of symptoms in schizophrenia (Kelly et al., 2011). Many of these investigative indications have been inspired by the subjective effect profile of cannabis and THC. These subjective effects, which are largely attributable to THC, can be quite diverse. The main effect is a “high” feeling, which is described by Ashton (2001, p. 104) as “a feeling of intoxication, with decreased anxiety, alertness, depression and tension and increased sociability”. Less frequently, there can also be effects that are reminiscent of psychosis, like distorted perceptions of colour, space and time. Other effects are impairments of reaction time, short‐term memory and motor coordination. Dysphoric reactions, described by Ashton (2001) as “anxiety and panic, paranoia and psychosis” can also occur. In addition, cannabis can induce feelings of appetite (“munchies”). The factors that determine the intensity of each of these effects have not been investigated in detail, but the subjective effects of THC seem to differ among users, and probably also between occasions of use or doses and modes of administration. There are many ways in which the (subjective) effects of cannabis and THC can be measured. In a review by Zuurman et al. (2009) feeling high was shown to be the most sensitive central nervous system (CNS) biomarker for the effects of cannabis, essentially irrespective of how this was measured. A frequently used tool to measure subjective feelings is the visual analogue scale (VAS). A VAS typically consists of a 100 mm long line, with two extremes on the sides. A subject is asked to indicate his or her current feelings somewhere on the line between the two extremes. Bowdle et al. (1998) described a composite scale for psychedelic effects (hereafter VAS Bowdle, see Table 1), consisting of 13 questions with the extremes of “not at all” and “extremely”, and validated this scale in a group of healthy volunteers who received ketamine. Zuurman et al. (2008) used the VAS Bowdle to measure the subjective effects of THC. Based on cluster analysis and factor analysis, they suggest the use of two distinct composite scales, which were classified as “internal perception” (VAS “reality”, “voices”, “meaning”, “suspicious” and “anxious”) and “external perception” (VAS “body”, “surroundings”, “time”, “thoughts”, “colours” and “sounds”), in addition to the item “high” (Zuurman et al., 2008). This same study also reported dose‐related effects on subjective alertness, which were assessed using the VAS described by Bond and Lader (1974). These authors identified 16 combinations of two subjective states (Table 1), combined in clusters of alertness (VAS “drowsy”, “feeble”, “muzzy”, “clumsy”, “lethargic”, “mentally slow”, “dreamy”, “incompetent” and “bored”), calmness (VAS “calm” and “relaxed”) and mood (VAS “contented”, “tranquil”, “happy”, “amicable” and “gregarious”), based on a principal component analysis (PCA) on response in a group of healthy volunteers, without intervention. Norris (1971) had previously subdivided these 16 items into four categories of four items each, based on a conceptual framework. These categories are “mental sedation or intellectual impairment” (items “drowsy”, “muzzy”, “mentally slow” and “dreamy”), “physical sedation or bodily impairments” (items “feeble”, “clumsy”, “lethargic” and “incompetent”), “tranquillization or calming effects” (items “calm”, “contented”, “tranquil” and “relaxed”) and “other types of feelings or attitudes” (items “happy”, “amicable”, “bored” and “gregarious”).

Table 1.

Description of visual analogue scales (VASs)

VAS Bowdle
Item Name Full description
1 Body My body or body parts seemed to change their shape or position.
2 Surroundings My surroundings seemed to change in size, depth, or shape.
3 Time The passing of time was altered.
4 Reality I had feelings of unreality.
5 Thoughts It was difficult to control my thoughts.
6 Colours The intensity of colours changed.
7 Sound The intensity of sound changed.
8 Voices I heard voices or sounds that were not real.
9 Meaning I had the idea that events, objects, or other people had particular meaning that was specific for me.
10 Suspicious I had suspicious ideas or the belief that others were against me.
11 High I felt high
12 Drowsy I felt drowsy.
13 Anxious I felt anxious.
VAS Bond and Lader
Item Name First extreme Second extreme
1 Drowsy Alert Drowsy
2 Calm Calm Excited
3 Feeble Strong Feeble
4 Muzzy Muzzy Clear‐headed
5 Clumsy Well‐coordinated Clumsy
6 Lethargic Lethargic Energetic
7 Contented Contented Discontented
8 Tranquil Troubled Tranquil
9 Mentally slow Mentally slow Quick witted
10 Relaxed Tense Relaxed
11 Dreamy Attentive Dreamy
12 Incompetent Incompetent Proficient
13 Happy Happy Sad
14 Amicable Antagonistic Amicable
15 Bored Interested Bored
16 Gregarious Withdrawn Gregarious
Open in a new tab

In combination, the 13 VAS described by Bowdle and the 16 scales used by Bond and Lader cover most of the subjective effects of cannabis and THC that were summarized by Ashton (2001), with the exception of the effects on appetite. The subjective effects of cannabis and THC are relevant, considering their putative therapeutic potential and pathophysiological significance. The aim of the current analysis is to identify distinct profiles within the spectrum of characteristic subjective effects of THC as measured using well‐known sets of VAS. Such distinct effect profiles could provide quantitative information on different neurophysiological effects of THC, and on different sensitivities of individuals to such effects. When these profiles are combined in composite scales, they can be used in the design and interpretation of studies assessing the effect of THC or cannabis and to improve our understanding of the endocannabinoid system in health and disease. For example, the relation between these subjective effects and personality or genetic constitution could be examined (van Winkel and Genetic Risk and Outcome of Psychosis (GROUP) Investigators, 2011), or the relationship between activation of certain brain regions in neuroimaging studies and subjective response patterns of THC or cannabis (e.g. Atakan et al., 2013). Individual VAS items are compared on sensitivity to the effects of THC, including a possible dose–response relationship. Also, different multivariate techniques were employed to examine if the clustering of different VAS items elicited distinct response patterns.

Methods

Data collection

Data from 10 studies performed by the Centre for Human Drug Research (CHDR) in which THC was administered to a total of 217 healthy volunteers were selected to perform an exploratory analysis on the measurements of the subjective effects of THC. The time points and measurements of VAS Bond and Lader and VAS Bowdle, as well as the time points of drug administration and administered dose were used for the analysis. All the studies had a randomized, cross‐over, placebo‐controlled design and were approved by the local Ethics Committee. Some studies were interaction studies, but only the treatment arms that involved administration of either THC alone or placebo alone were taken into account. An overview of the studies and their references is provided in Table 2. All healthy volunteers who participated in the studies were mild cannabis users, defined as a frequency of cannabis use of maximum once a week in the past year.

Table 2.

Overview of original studies

Reference N Doses Interval VAS timepoints

Bossong et al., 2009

7 8 mg inhalation Not available 7, 12, 17, 32 and 105 minutes after dose

van Hell et al., 2011

26 6 + 1 + 1 + 1 mg 30 minutes 27 and 34 minutes after first dose

Kleinloog et al., 2012

49 2 + 4 + 6 mg 90 minutes 13, 25, 33 and 64 minutes after each dose

Klumpers et al., 2012a

12 5, 6.5 or 8 mg Oral 19, 36, 51, 65, 95 minutes after dose

Klumpers et al., 2012b

22 2 + 6 + 6 mg 90 minutes 29, 59 and 83 minutes after each dose

Klumpers et al., 2013a

30 2 + 4 + 6 + 6 mg 60 minutes 23 and 41 minutes after each dose
Klumpers et al., 2013b 34 5 × 4 mg ≥ 150 minutes 10, 24 and 115 minutes after each dose

Liem‐Moolenaar et al., 2010

37 2 + 4 + 6 mg 90 minutes 22, 34 and 61 minutes after each dose

Zuurman et al., 2008

12 2 + 4 + 6 + 8 mg 90 minutes 22 and 47 minutes after each dose

Zuurman et al., 2010

36 2 + 4 + 6 + 6 mg 60 minutes 23 and 41 minutes after each dose
Open in a new tab

THC challenge

In nine out of 10 studies (94.5% of subjects) purified THC was inhaled using the Volcano™ vaporizer (Storz‐Bickel, Tuttlingen, Germany). This method is described in more detail by Zuurman et al. (2008). In all these studies, several administrations were given during a study day to prolong the effect of THC. This was typically an increasing dose (2 mg, 4 mg, 6 mg) with 60 to 90 minute intervals, although the actual dosing regimen was different throughout the studies (see Table 2). One study used single doses of oral or sublingual tablets of purified THC. This study was included as a check of the notion that THC effects are determined by individual sensitivity and brain concentrations, and not by administration route.

Quality control

Prior to the analysis, a visual quality check of the available data was performed. In this regard, for each VAS item the data of the placebo condition from different studies were presented as boxplots. As no subjective effects are expected during the placebo condition, the scores for the items of the VAS Bond and Lader were expected to be distributed around the middle and the scores for the items of the VAS Bowdle close to a score of 0 mm. A few studies showed a distribution during the placebo condition that was distinctly different from the other studies (based on visual comparison), and these were excluded from further analysis.

Item sensitivity

Measurements performed in the first 60 minutes after THC administration were pooled to identify items that are sensitive to THC. Items that showed a significant difference between THC and placebo were selected for further analysis. A Kruskal–Wallis test was used as the distribution was not normal. A p‐value of 0.05/29 (Bonferroni correction for number of VAS items) was considered significant.

Defining responders

Not all subjects showed a response on the VAS after administration of THC. Subjects were therefore classified as responder or non‐responder for each individual VAS item. To make this classification, the distribution of observed scores for the overall placebo condition was examined, and the values within the 95% observation interval during placebo were considered indicative for the absence of a response. Conversely, subjects were considered a responder for a specific VAS item if they showed a response outside this 95% limit during any measurement in the THC condition.

Dose–response relationships

To determine possible dose–response relationships for the different VAS items that are sensitive for the effect of THC, the studies that used intrapulmonary administration of THC were selected. Most of these studies had a design where multiple doses of THC were administered on each study day, with a fixed time interval between administrations. As the times of measurements were different for each study, the maximum response after each administration was used. All individual studies were designed to include measurements around the expected maximum effect (T max). Possible relationships were tested using a Kruskal–Wallis test and a p‐value of 0.05/n (Bonferroni correction) was considered significant. Since most studies included several different consecutive doses on each individual study occasion, it was possible to assess a dose–response relationship, which was performed in two steps. Initially, only the first administration of THC during each study day was taken into account, which assured the absence of carry‐over effects and tolerance. Subsequently, all administrations during each study day were considered if the dosing interval was at least 60 minutes, which covered more observations and a larger dose range. Both steps were repeated within subjects who were identified as responders.

Cluster selection

For the items that showed a significant dose–response relationship, different methods were applied to determine the combination of (weighted) items that could best describe “the subjective effect of THC”. The combinations of items found with these different methods were then compared on their ability to predict the drug condition (THC or placebo). To find clusters within the dataset, multiple correspondence analysis (MCA), PCA, factor analysis (FA), k‐means cluster analysis (KCA), hierarchical cluster analysis (HCA), variable clustering (VC) and discriminant analysis (DA) were used. Each technique has its own advantages and should more or less lead to the same conclusion if the clusters are the result of an underlying construct. The final cluster selection was based on what the different clustering methods have in common. MCA is an exploratory technique that uses logical indicators (true or false), which makes it more suitable for data that is not normally distributed or categorical (Greenacre, 2007). For MCA, the dataset was recoded into responders and non‐responders, and the final item selection was based on the inertia of the items. PCA is the most commonly used tool in exploratory data analysis (Jolliffe, 2002). FA is a technique similar to PCA, but it only focuses on the variability that is shared with another item, whereas PCA takes all variability into account (Jolliffe, 2002). Both techniques can be applied to a dataset that is jointly normally distributed and are sensitive to the relative scaling of the original variables. PCA and FA were therefore performed using the maximum response in the THC condition, after a mean subtraction for all VAS items and log‐transformation for VAS Bowdle items. KCA is a disjoint clustering method, in which all items are distributed within a pre‐defined number of separated clusters to minimize within‐cluster variability and maximize between‐cluster variability (Hartigan and Wong, 1979). Within HCA and VC, clusters are organized to identify a hierarchical structure based on similarity between items, which is typically presented as a dendrogram (Jain et al., 1999). Linear stepwise DA is another method to find combinations of items that are able to predict the subjective effect of THC. For this analysis, the maximum response following intrapulmonary administration of THC was used to select items and the data from the study that was performed most recently were exclusively used for cross‐validation. A stopping criterion of 0.1% improvement was used for forward and combined analysis and a stopping criterion of −0.5% improvement (either any improvement or a maximum of 0.5% worsening) for backward analysis.

Inverse predictive check

An inverse predictive check was performed for the individual items and the possible combinations of items to compare the probability to identify the original treatment. In this regard, predictive values were calculated for each individual VAS item, the clusters described by Bond and Lader (1974) and Zuurman et al. (2008), and the combinations of items found in the current analysis. The predictive value describes the chance that the score on a certain item (or combination of items) correctly identifies the given treatment (THC or placebo). As there are two possible outcomes (THC or placebo), the a priori predictive value is 50%. Again, the data from the study that was most recently performed were exclusively used for cross‐validation.

Statistical software

The open source statistical software package R (version 2.14.0, http://www.r-project.org) was used for the analyses.

Results

Quality control

Based on a visual check of the distribution of scores under placebo conditions and prior to other evaluations, three studies [with a total of 74 (34.1%) subjects] were excluded from further analysis, based on the scores under placebo on all VAS items. One study showed a slightly different placebo profile compared with the other studies, which was a positron emission tomography (PET)‐study with administration of [11C]‐raclopride and PET‐measurements during THC administration in both study arms. It was decided to include the information from this study with the use of placebo correction.

Item sensitivity

VAS Bond and Lader items “contented”, “tranquil”, “happy”, “amicable” and “bored” were the only items that did not show a statistically significant different score between THC and placebo conditions (after Bonferroni correction). All these items are part of Bond and Lader's “mood” cluster, with the exception of VAS “bored”.

Selection of responders

Table 2 shows the upper and lower limits of the 95% observation interval for VAS Bowdle and VAS Bond and Lader, respectively, during all placebo occasions. Individual scores outside of these limits during THC occasions were considered to be indicative of a drug response. The percentages of subjects who were classified as item responders after THC administration are presented per item.

Dose–response relationships

When taking into account all administrations and all subjects, VAS “drowsy”, “feeble”, “clumsy” and “dreamy” of Bond and Lader and all VAS Bowdle items except “voices” and “anxious” showed a significant dose − response relation using a p‐value of 0.05/96 (Bonferroni correction for four times 24 items; the five items that did not differ significantly between placebo and THC were not taken along). The items that showed a significant dose − response relation for all administrations and all subjects were used for cluster selection. It should be noted that the dose interval (mostly between 60 and 90 minutes) is likely to have resulted in an accumulation of effect. The dose level of 8 mg is not included in the dose − response analysis, as only eight observations were available for this dose (compared to 150, 110, 118 and 101 observations for the other doses).

Multiple correspondence analysis

The outcome of the MCA is presented in Figure 1. The closer two data points are to one another, the more likely they are to show a response at the same time. Items that are relatively closer to the right side of the map are items that are more likely to show a response than items that are closer to the left side of the map. The relative inertia of the items is provided in Table 2. Inertia is a measure of how much the item contributes independent of other items (comparable to eigenvalues).

Figure 1.

Figure 1

Open in a new tab

Map of multiple correspondence analysis (MCA). The final cluster selection based on all methods has been highlighted (red squares: dysphoria; blue triangles: perception; green circles: relaxation).

Principal component analysis (PCA) and factor analysis (FA)

Following parallel analysis, two components were selected for the varimax rotated PCA based on the observation of a “sharp break” in the scree plot. Another way of determining the optimal number of components is comparing the eigenvalues of the possible components in the dataset with those obtained from a random, simulated dataset of the same size. Using this approach, three components would have been selected. Figure 2 presents a map of the rotated PCA and the factor loadings are presented in Table 2. For FA (based on maximum likelihood), parallel analysis suggested the use of five factors, which are presented in Figure 3.

Figure 2.

Figure 2

Open in a new tab

Plot of rotated principal component analysis (PCA) (green: relaxation; red: dysphoria; blue: perception).

Figure 3.

Figure 3

Open in a new tab

Overview of factor analysis (FA) (green: relaxation; red: dysphoria; blue: perception).

Cluster analysis

Within KCA, all items were assigned to one of three clusters. This technique gives no indication as to how well the variable fits into the cluster. Although items were scaled to allow for better comparison, all items of VAS Bond and Lader grouped into one cluster and VAS Bowdle items “surroundings”, “colours”, “sound” and “suspicious” were separated from the remaining items. The results of HCA are presented in Figure 4. Variable clustering had similar results. Because all these clustering methods will place all items within a cluster, the items that do not cluster consistently throughout the methods (e.g. item “drowsy” of VAS Bowdle) are likely irrelevant.

Figure 4.

Figure 4

Open in a new tab

Overview of hierarchical cluster analysis (HCA). The final cluster selection based on all methods has been highlighted (green: relaxation; red: dysphoria; blue: perception).

Discriminant analysis (DA)

Forward, backward and combined discriminant analyses were performed. The results from the combined DA were equal to the results of forward DA and are therefore not presented separately. Forward DA identified VAS “calm”, “dreamy”, “incompetent” and “high” as most predictive for the effect of THC and backward DA identified VAS “lethargic”, “relaxed”, “incompetent”, “bored”, “gregarious”, “thoughts” and “high”. As VAS “high” was expected to have a large impact on the outcomes, the analysis was repeated without this item, resulting in a combination of VAS “lethargic”, “dreamy”, “thoughts” and “colours” in case of forward DA and the same items together with VAS “sound” in case of backward DA.

Inverse predictive check

To calculate the predictive value of the different combinations of items, two methods were used: a composite (average) score of the items and a combination of the individual scores on the different items. Individually, VAS “high” has the best predictive value (83.6%), followed by VAS “thoughts” (77.1%), VAS “mentally slow” (75.8%) and VAS “time” (75.2%), as presented in Table 2. From the different combinations of items, only those found with DA resulted in a better predictive value as a composite scale.

Discussion

THC and cannabis have a rather broad range of effects, which can differ between subjects, doses and use circumstances. The effect patterns can give insight into the many different functions, therapeutic areas and diseases in which the cannabinoid system has been implicated. This analysis explored the characteristics of the scope of subjective effects of THC as measured on different VASs. As THC is the main component of cannabis and the elicited subjective effects of THC are comparable to the subjective effects of cannabis described in the literature, the findings might be applicable to cannabis. However, the other components of cannabis (i.e. cannabidiol) might distort the subjective effect patterns. It is important to note that different preparations of cannabis have different levels of THC and cannabidiol. The analysis examines the effects of THC in mild cannabis users. Results might be relevant to other groups of people (heavy users, non‐users), although further research is needed.

VAS Bond and Lader and VAS Bowdle are frequently used in CNS drug research and capture most of the subjective effects that have been described with cannabis or THC. Not surprisingly, the analysis indicated that feeling high was the most predictive item for the effects of THC, which confirms the literature review of Zuurman et al. (2009) that showed a statistically significant “high” effect in 96% of cannabis studies. The other items that had high individual predictive values describe effects on time perception and cognitive functions (controlling of thoughts and mental slowness), which are also well known and frequent effects of cannabis and THC (Ashton, 2001). The items with high predictive values could be grouped into three distinct factors of effect. Table 3 presents an overview of the proposed composite scale that measures these factors.

Table 3.

Overview of different outcome measures

VAS item LLOI ULOI % Responders % Responders (0 mg) % Responders (2 mg) % Responders (4 mg) % Responders (6 mg) MCA: % inertia PCA: comp. 1 PCA: comp. 2 Predictive value
Drowsy 36 61 62 13 21 37 52a 5.30 0.80 0.20 70.8
Calm 35 54 56 7 8 11 18 2.89 0.28 0.41 57.6
Feeble 37 56 52 7 17 30 45a 5.56 0.82 0.27 68.3
Muzzy 44 64 69 9 20 26 51 5.01 0.85 0.29 74.4
Clumsy 38 53 69 12 34 47 56a 5.25 0.80 0.30 71.5
Lethargic 35 62 66 7 9 16 20 0.60 ‐0.15 0.11 47.7
Contented 34 53 48 35 44 58 61 3.54 0.42 0.16 63.6
Tranquil 48 66 68 19 19 25 36 1.27 0.37 0.44 71.6
Mentally slow 42 64 69 9 22 36 54 5.53 0.84 0.23 75.8
Relaxed 46 66 53 11 15 21 34 2.13 0.52 0.12 72.4
Dreamy 38 61 64 5 23 41 55a 5.31 0.82 0.26 72.3
Incompetent 48 64 62 13 12 16 31 2.65 0.73 0.37 66.4
Happy 32 52 49 7 15 13 32 2.91 0.59 0.33 56.3
Amicable 48 68 47 20 26 20 27 0.59 0.47 0.09 47.6
Bored 37 62 34 6 6 14 18 1.26 0.68 0.04 53.8
Gregarious 47 69 50 16 17 15 30 1.34 0.68 0.38 61.8
Body 0 1 42 5 15 24 31a 3.71 0.26 0.69 55.9
Surroundings 0 1 40 1 15 25 37a 4.60 0.37 0.76 65.1
Time 0 1 63 4 27 48 57a 5.00 0.49 0.65 75.2
Reality 0 1 48 7 24 34 46a 4.63 0.37 0.77 67.6
Thoughts 0 1 74 7 39 63 66a 5.30 0.55 0.60 77.1
Colours 0 1 46 3 19 29 42a 3.76 0.27 0.69 63.9
Sound 0 1 49 3 20 31 39a 4.52 0.46 0.66 68.5
Voices 0 1 25 3 7 14 19 1.91 ‐0.01 0.84 54.2
Meaning 0 1 22 2 7 10 17a 1.71 0.09 0.76 53.8
Suspicious 0 1 21 4 7 12 16a 1.50 0.01 0.84 52.8
High 0 2 88 1 58 80 87a 4.63 0.58 0.46 83.6
Drowsy 0 3 58 7 30 40 61a 4.63 0.61 0.46 66.6
Anxious 0 1 35 4 11 19 30 2.95 0.29 0.78 56.8
Open in a new tab
a

Significant dose − response relationship.

The first common factor that was found using the different methods of cluster selection consists of VAS “time”, “thoughts” and “high”. VAS “colours” and “sound” showed a relation with this factor. Together, this factor can be described conceptually as a measure of feeling high and changes in perception (“perception”). All these items were a part of the cluster “external perception” as described by Zuurman et al. (2008), except “high” which was treated by Zuurman et al. (2008) as a separate cluster because of its predominance. This clustering of feeling high and the other items follows the description of the most typical THC effects by Ashton (2001).

VAS “drowsy” (from VAS Bond and Lader), “muzzy”, “mentally slow” and “dreamy” constitute the second common factor. VAS “feeble” and “clumsy” showed a relation with this factor. All these items are included in the “alertness” clusters as described by Bond and Lader (1974). The main four items can be seen as mental aspects of sedation (“relaxation”), whereas the two related items are more physical aspects of sedation.

VAS “voices”, “meaning” and “suspicious” are included in the third common factor within the effects of THC. These items may represent what Ashton (2001) describes as “dysphoric reactions” (“dysphoria”). Zuurman et al. (2008) included these items in the “internal perception” cluster. Not many subjects show an effect on these items, but if these scales are affected, the effects seem to be clear. Even though this cluster of items does not seem highly predictive of the effect of THC, this aspect might correlate with other important predictors for the effect of THC such as the occurrence of adverse events. Henquet et al. (2005) suggest these “dysphoric reactions” could also reflect a predisposition for the development of psychosis.

Within the PCA, these three factors (“perception”, “relaxation” and “dysphoria”) were shown to represent two principal components. The majority of the variation was explained by “relaxation” (aligned with the horizontal component in Figure 2). The remaining variation could be explained by “dysphoria” (aligned with the vertical component in Figure 2), which appears to be the opposite of mental relaxation. The “perception” effects (the more “typical” effects of THC) were not a part of the two components, but rather seem to represent a separate component that is a vector of the other two clusters and therefore represents the main underlying effect. As described earlier, another way of determining the number of components would have resulted in three components.

The items that did not show a significant difference between THC and placebo are a part of the “mood” cluster in the VAS Bond and Lader, with the exception of VAS “bored” (a scale that does seem to relate to mood). This would suggest that THC does not affect mood in a stricter sense, which is consistent with the review by Zuurman et al. (2009). The effect of THC on appetite is not measured by VAS Bond and Lader or VAS Bowdle and therefore not taken along in the current analysis. Given the relevance of THC (and the endocannabinoid system) on appetite (Farrimond et al., 2011), it would have been interesting to observe how additional VAS scales of hunger and appetite would have behaved in relation with the other clusters. For use in future studies, the authors would recommend the addition of scales for hunger and appetite.

A more complete assessment of the different effect dimensions of THC could aid in the exploration of the various pharmacological and physiologic aspects of cannabinoid systems, in health and disease. For example, the interactions of different constituents of cannabis (i.e. THC and cannabidiol) could be disentangled, by measuring the effects of each component and different combinations of components (Bhattacharyya et al., 2010). Also, the effect profiles can help in quantifying the dose − response relationships for different THC‐effects, for instance to discriminate peripheral and central cannabinoid type 1 antagonists (Klumpers et al., 2013b). Other applications of the dimensional scales could be in exploring the brain structures underlying different effects of THC (Atakan et al., 2013), as well as examining the influence of genetic factors (i.e. polymorphisms) on the subjective effects of THC or cannabis (van Winkel and Genetic Risk and Outcome of Psychosis (GROUP) Investigators, 2011). The dimensional quantification of cannabinoid effects can also help in exploring the relation between subjective effects to cannabis and clinical risk of psychosis, for instance by demonstrating that patients (or people at risk) show relatively strong “perception” effects, compared to the other effects of a cannabinoid challenge (Henquet et al., 2010).

The skewed distribution of VAS Bowdle is unfavourable for statistical analysis. The finding that a combination of items of the VAS Bond and Lader, which has an approximately normal distribution, explains most of the variation in effect is therefore important. The clear separation between items from VAS Bond and Lader and from VAS Bowdle that was seen with most methods is interesting. This could be caused by the differences in distribution that are characteristic for the scales, which is the result of (1) the use of two‐sided versus one‐sided scales and (2) the use of effects that are present and absent under “normal” circumstances (i.e. it is normal to have fluctuations in mood and alertness, but not in psychedelic effects). However, the separation could also be caused because the psychometric properties of the scales (i.e. what they measure) are different.

In summary, the current analysis provides experimental evidence that the subjective effects of THC in mild cannabis users have three main dimensions (consistent over a variety of statistical techniques). The main subjective effects of THC consist of feeling high and changes in perception. In addition, mental relaxation or dysphoric reactions can occur more or less independently. These findings correspond with previous descriptions of the subjective effects of THC and cannabis. The three dimensions can be used as the basis of an evidence‐based composite scale (see Table 4, which could also include effects on hunger and appetite), to further explore and differentiate the involvement of endocannabinoid systems in health and disease and to quantify the subjective effects of THC and cannabis in clinical research. There seems to be a subset of individuals (even among occasional cannabis users) who respond to THC with dysphoric reactions and another small group of individuals who do not experience the typical “high” effects of THC. Further exploration of the genetic or psychological profiles of these individuals and the relation with subjective effect patterns could shed more light on the role of the cannabinoid system in health and (mental) disease.

Table 4.

Overview of suggested composite scale

Subscale perception
Time perception
Change in control of thoughts
Feeling high
Subscale relaxation
Feeling drowsy
Feeling muzzy, not having a clear head
Mental slowness
Feeling dreamy
Subscale dysphoria
Hearing voices
The idea that events, objects or people have a special meaning
Suspicious ideas or beliefs
Subscale appetite
Feelings of hunger
Feelings of appetite
Open in a new tab

Declaration of interest statement

The authors declare no conflict of interest.

Acknowledgements

The authors would like to thank Marieke Liem‐Moolenaar and Linda Klumpers for their support in data collection. This research was paid for by the Centre for Human Drug Research.

References

  1. Ashton C.H. (2001) Pharmacology and effects of cannabis: a brief review. The British Journal of Psychiatry, 178(2), 101–6. DOI: 10.1192/bjp.178.2.101 [DOI] [PubMed] [Google Scholar]
  2. Atakan Z., Bhattacharyya S., Allen P., Martín‐Santos R., Crippa J.A., Borgwardt S.J., Fusar‐Poli P., Seal M., Sallis H., Stahl D., Zuardi A.W., Rubia K., McGuire P. (2013) Cannabis affects people differently: inter‐subject variation in the psychotogenic effects of Δ9‐tetrahydrocannabinol: a functional magnetic resonance imaging study with healthy volunteers. Psychological Medicine 43(6), 1255–67. DOI: 10.1017/S0033291712001924 [DOI] [PubMed] [Google Scholar]
  3. Berry E.M., Mechoulam R. (2002) Tetrahydrocannabinol and endocannabinoids in feeding and appetite. Pharmacology Therapy, 95(2), 185–190. DOI: 10.1016/S0163-7258(02)00257-7 [DOI] [PubMed] [Google Scholar]
  4. Bhattacharyya S., Morrison P.D., Fusar‐Poli P., Martin‐Santos R., Borgwardt S., Winton‐Brown T., Nosarti C., O'Carroll C.M., Seal M., Allen P., Mehta M.A., Stone J.M., Tunstall N., Giampietro V., Kapur S., Murray R.M., Zuardi A.W., Crippa J.A., Atakan Z., McGuire P.K. (2010) Opposite effects of delta‐9‐tetrahydrocannabinol and cannabidiol on human brain function and psychopathology. Neuropsychopharmacology, 35(3), 764–774. DOI: 10.1038/npp.2009.184 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bond A., Lader M. (1974) Use of analog scales in rating subjective feelings. British Journal of Medical Psychology, 47: 211–218. DOI: 10.1111/j.2044-8341.1974.tb02285.x [DOI] [Google Scholar]
  6. Bossong M.G., van Berckel B.N., Boellaard R., Zuurman L., Schuit R.C., Windhorst A.D., van Gerven J.M., Ramsey N.F., Lammertsma A.A., Kahn R.S. (2009) ∆ 9‐Tetrahydrocannabinol induces dopamine release in the human striatum. Neuropsychopharmacology, 34(3), 759–766. DOI: 10.1038/npp.2008.138 [DOI] [PubMed] [Google Scholar]
  7. Bowdle T.A., Radant A.D., Cowley D.S., Kharasch E.D., Roy‐Byrne P.P. (1998) Psychedelic effects of ketamine in healthy volunteers: relationship to steady‐state plasma concentrations. Anesthesiology, 88(1), 82–88. DOI: 10.1038/sj.npp.1395214 [DOI] [PubMed] [Google Scholar]
  8. Compton D.R., Aceto M.D., Lowe J., Martin B.R. (1996) In vivo characterization of a specific cannabinoid receptor antagonist (SR141716A): inhibition of delta 9‐tetrahydrocannabinol‐induced responses and apparent agonist activity. Journal of Pharmacology and Experimental Therapeutics, 277(2), 586–594. [PubMed] [Google Scholar]
  9. D'Souza D.C., Perry E., MacDougall L., Ammerman Y., Cooper T., Wu Y.T., Braley G., Gueorguieva R., Krystal J.H. (2004) The psychotomimetic effects of intravenous Δ9‐tetrahydrocannabinol in healthy individuals: implications for psychosis. Neuropsychopharmacology, 29(8), 1558–1572. DOI: 10.1038/sj.npp.1300496 [DOI] [PubMed] [Google Scholar]
  10. Farrimond J.A., Mercier M.S., Whalley B.J., Williams C.M. (2011) Cannabis sativa and the endogenous cannabinoid system: therapeutic potential for appetite regulation. Phytotherapy Research, 25(2), 170–188. DOI: 10.1002/ptr.3375 [DOI] [PubMed] [Google Scholar]
  11. van Gaal L.F., Rissanen A.M., Scheen A.J., Ziegler O., Rössner S. (2005) Effects of the cannabinoid‐1 receptor blocker rimonabant on weight reduction and cardiovascular risk factors in overweight patients: 1‐year experience from the RIO‐Europe study. Lancet 365(9468), 1389–1397. DOI: 10.1016/S0140-6736(05)66374-X [DOI] [PubMed] [Google Scholar]
  12. Greenacre M. (2007) Correspondence Analysis in Practice, 2nd edn, Boca Raton, FL, Chapman & Hall/CRC. [Google Scholar]
  13. Hartigan J.A., Wong, M.A. (1979) Algorithm AS 136: a k‐means clustering algorithm. Applied Statistics, 28(1), 100–108. DOI: 10.2307/2346830 [DOI] [Google Scholar]
  14. van Hell H.H., Bossong M.G., Jager G., Kristo G., van Osch M.J., Zelaya F., Kahn R.S., Ramsey N.F. (2011) Evidence for involvement of the insula in the psychotropic effects of THC in humans: a double‐blind, randomized pharmacological MRI study. International Journal of Neuropsychopharmacology, 14(10), 1377–1388. DOI: 10.1017/S1461145711000526 [DOI] [PubMed] [Google Scholar]
  15. Henquet C., Krabbendam L., Spauwen J., Kaplan C., Lieb R., Wittchen H.U., van Os J. (2005) Prospective cohort study of cannabis use, predisposition for psychosis, and psychotic symptoms in young people. BMJ, 330(7481), 11 DOI: 10.1136/bmj.38267.664086.63 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Henquet C., van Os J., Kuepper R., Delespaul P., Smits M., à Campo J., Myin‐Germeys I. (2010) Psychosis reactivity to cannabis use in daily life: an experience sampling study. British Journal of Psychiatry, 196, 447–453. DOI: 10.1192/bjp.bp.109.072249 [DOI] [PubMed] [Google Scholar]
  17. Jain A.K., Murty M.N., Flynn P.J. (1999) Data clustering: a review. ACM Computing Surveys, 31(3), 264–323. DOI: 10.1145/331499.331504 [DOI] [Google Scholar]
  18. Jolliffe I.T. (2002) Principal Component Analysis, 2nd edn, New York, Springer. [Google Scholar]
  19. Karst M., Wipperman S., Ahrens J. (2010) Role of cannabinoids in the treatment of pain and (painful) spasticity. Drugs, 70(18), 2409–2438. [DOI] [PubMed] [Google Scholar]
  20. Kelly D.L., Gorelick D.A., Conley R.R., Boggs D.L., Linthicum J., Liu F., Feldman S., Ball M.P., Wehring H.J., McMahon R.P., Huestis M.A., Heishman S.J., Warren K.R., Buchanan R.W. (2011) Effects of the cannabinoid‐1 receptor antagonist rimonabant on psychiatric symptoms in overweight people with schizophrenia: a randomized, double‐blind, pilot study. Journal of Clinical Psychopharmacology, 31(1), 86–91. DOI: 10.1097/JCP.0b013e318204825b [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Kleinloog D., Liem‐Moolenaar M., Jacobs G., Klaassen E., de Kam M.L., Hijman R., van Gerven J. (2012) Does olanzapine inhibit the psychomimetic effects of Δ9‐tetrahydrocannabinol? Journal of Psychopharmacology, 26(10), 1307–1316. DOI: 10.1177/0269881112446534 [DOI] [PubMed] [Google Scholar]
  22. Klumpers L.E., Beumer T.L., van Hasselt J.G., Lipplaa A., Karger L.B., Kleinloog H.D., Freijer J.I., de Kam M.L., van Gerven J.M. (2012a) Novel Δ(9)‐tetrahydrocannabinol formulation Namisol® has beneficial pharmacokinetics and promising pharmacodynamic effects. British Journal of Clinical Pharmacology, 74(1), 42–53. DOI: 10.1111/j.1365-2125.2011.04164.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Klumpers L.E., Cole D.M., Khalili‐Mahani N., Soeter R.P., Rombouts S.A., van Gerven J.M. (2012b) Manipulating brain connectivity with δ9‐tetrahydrocannabinol: a pharmacological resting state FMRI study. NeuroImage 63(3), 1701–1711. DOI: 10.1016/j.neuroimage.2012.07.051 [DOI] [PubMed] [Google Scholar]
  24. Klumpers L.E., Roy C., Ferron G., Turpault S., Poitiers F., Pinquier J.L., van Hasselt J.G., Zuurman L., Erwich F.A., van Gerven J.M. (2013a) Surinabant, a selective CB(1) antagonist, inhibits THC‐induced central nervous system and heart rate effects in humans. British Journal of Clinical Pharmacology, 76(1), 65−77. DOI: 10.1111/bcp.12071. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Klumpers L.E., Fridberg M., de Kam M.L., Little P.B., Jensen N.O., Kleinloog H.D., Elling C.E. (2013b) Peripheral selectivity of the novel cannabinoid receptor antagonist TM38837 in healthy subjects. British Journal of Clinical Pharmacology, 76(6), 846–857. DOI: 10.1111/bcp.12141 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Liem‐Moolenaar M., te Beek E.T., de Kam M.L., Franson K.L., Kahn R.S., Hijman R., Touw D., van Gerven J.M. (2010) Central nervous system effects of haloperidol on THC in healthy male volunteers. Journal of Psychopharmacology, 24(11), 1697–1708. DOI: 10.1017/S1461145711000526 [DOI] [PubMed] [Google Scholar]
  27. Lynch M.E., Campbell F. (2011) Cannabinoids for treatment of chronic non‐cancer pain; a systematic review of randomized trials. British Journal of Clinical Pharmacology, 72(5), 735–744. DOI: 10.1111/j.1365-2125.2011.03970.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Norris H. (1971) The action of sedatives on brain stem oculomotor systems in man. Neuropharmacology, 10(2), 181–191. DOI: 10.1016/0028-3908(71)90039-6 [DOI] [PubMed] [Google Scholar]
  29. van Oosten B.W., Killestein J., Mathus‐Vliegen E.M.H., Polman C.H. (2004) Multiple sclerosis following treatment with a cannabinoid receptor‐1 antagonist. Multiple Sclerosis 10(3), 330–2. DOI: 10.1016/S0140-6736(05)66374-X [DOI] [PubMed] [Google Scholar]
  30. van Winkel R., Genetic Risk and Outcome of Psychosis (GROUP) Investigators . (2011) Family‐based analysis of genetic variation underlying psychosis‐inducing effects of cannabis; sibling analysis and proband follow‐up. Archives of General Psychiatry, 68(2), 148–157. DOI: 10.1001/archgenpsychiatry.2010.152 [DOI] [PubMed] [Google Scholar]
  31. Zajicek J.P., Apostu V.I. (2011) Role of cannabinoids in multiple sclerosis. CNS Drugs, 25(3), 187–201. [DOI] [PubMed] [Google Scholar]
  32. Zuurman L., Roy C., Schoemaker R.C., Hazekamp A., den hartigh J., Bender J.C., Verpoorte R. Pinquier J.L., Cohen A.F., van Gerven J.M. (2008) Effect of intrapulmonary tetrahydrocannabinol administration in humans. Journal of Psychopharmacology, 22(7), 707–716, DOI: 10.1177/0269881108089870 [DOI] [PubMed] [Google Scholar]
  33. Zuurman L., Ippel A.E., Moin E., van Gerven J.M.A. (2009) Biomarkers for the effects of cannabis and THC in healthy volunteers. British Journal of Clinical Pharmacology, 67(1), 5–21. DOI: 10.1111/j.1365-2125.2008.03329.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Zuurman L., Roy C., Schoemaker R.C., Amatsaleh A., Guimaeres L., Pinquier J.L., Cohen A.F., van Gerven J.M. (2010) Inhibition of THC‐induced effects on the central nervous system and heart rate by a novel CB1 receptor antagonist AVE1625. Journal of Psychopharmacology, 24(3), 363–371. DOI: 10.1177/0269881108089870 [DOI] [PubMed] [Google Scholar]

Articles from International Journal of Methods in Psychiatric Research are provided here courtesy of Wiley

RESOURCES